In　this　paper,　an　evaluation　method　for　load　supply　capacity　based　on　evidence　theory　and　affine　analytic　recursion　is　proposed　to　address　the　limitations　of　existing　methods,　which　rely　on　initial　values　for　the　iteration　convergence　and　struggle　to　comprehensively　consider　the　objective　uncertainty　and　cognitive　uncertainty　information.　The　method　constructs　distributed　generation　output　model　using　evidence　theory　and　employs　the　step-varied　repeated　power　flow　algorithm　based　on　affine　analytic　recursion　to　calculate　the　focal　elements　of　the　maximum　load　growth　percentage.　The　obtained　results　accurately　represent　the　distribution　characteristics　of　the　maximum　load　growth　percentage,　thereby　establishing　confidence　intervals　for　evaluating　the　supply　risk　across　various　operating　conditions.　Cases　show　that,　the　proposed　method　is　effective　and　is　capable　of　significantly　reducing　computation　time.　©　2023　IEEE.